Thomas Gleixner <[email protected]> writes:
> Jiaxun Yang <[email protected]> writes:
>> +
>> +struct pch_msi_data {
>> + spinlock_t msi_map_lock;
>> + phys_addr_t doorbell;
>> + u32 irq_first; /* The vector number that MSIs
>> starts */
>> + u32 num_irqs; /* The number of vectors for
>> MSIs */
>> + unsigned long *msi_map;
>> +};
>> +
>> +static void pch_msi_mask_msi_irq(struct irq_data *d)
>> +{
>> + pci_msi_mask_irq(d);
>> + irq_chip_mask_parent(d);
>> +}
>> +
>> +static void pch_msi_unmask_msi_irq(struct irq_data *d)
>> +{
>> + pci_msi_unmask_irq(d);
>> + irq_chip_unmask_parent(d);
>
> The ordering of mask and unmask is assymetric. That does not make sense.
>
>> +static struct msi_domain_info pch_msi_domain_info = {
>> + .flags = MSI_FLAG_USE_DEF_DOM_OPS | MSI_FLAG_USE_DEF_CHIP_OPS |
>> + MSI_FLAG_MULTI_PCI_MSI | MSI_FLAG_PCI_MSIX,
>> + .chip = &pch_msi_irq_chip,
>
> Please maintain tabular layout.
Ooops. Wanted to reply to V3, but the comments are valid for V3 as well.
Thanks,
tglx
